Research on random fatigue damage analysis method considering the influence of mean stress

This paper proposes a frequency-domain fatigue damage evaluation method for broadband non-Gaussian random processes that accounts for mean stress effects. By incorporating a bandwidth parameter into the conditional mean probability density function, the fitting accuracy is improved. Furthermore, the relationships between skewness, kurtosis, S-N curve slope, and tensile strength are analyzed to derive a fatigue damage correction factor considering mean stress effects. Using rainflow-counting results corrected by the Goodman method in the time domain as a benchmark, numerical simulations are conducted with both parametric power spectra and real response spectra. Comparisons with existing non-Gaussian fatigue damage assessment methods demonstrate the accuracy and applicability of the proposed approach.